Exploring Factors Controlling the Variability of Pesticide Concentrations in the Willamette River Basin Using Tree-Based Models

Abstract

We analyzed available concentration data of five commonly used herbicides and three pesticides collected from small streams in the Willamette River Basin in Oregon to identify factors that affect the variation of their concentrations in the area. The emphasis of this paper is the innovative use of classification and regression tree models for exploratory data analysis as well as analyzing data with a substantial amount of left-censored values. Among variables included in this analysis, land-use pattern in the watershed is the most important for all but one (simazine) of the eight pesticides studied, followed by geographic location, intensity of agriculture activities in the watershed (represented by nutrient concentrations in the stream), and the size of the watershed. The significant difference between urban sites and agriculture sites is the variability of stream concentrations. While all 16 nonurban watersheds have significantly higher variation than urban sites, the same is not necessarily true for the mean concentrations. Seasonal variation accounts for only small fraction of the total variance in all eight pesticides.